A Role for Id2 in Regulating Photic Entrainment of the Mammalian Circadian System

Summary Inhibitor of DNA binding genes (Id1–Id4) encode helix-loop-helix (HLH) transcriptional repressors associated with development and tumorigenesis [1, 2], but little is known concerning the function(s) of these genes in normal adult animals. Id2 was identified in DNA microarray screens for rhythmically expressed genes [3–5], and further analysis revealed a circadian pattern of expression of all four Id genes in multiple tissues including the suprachiasmatic nucleus. To explore an in vivo function, we generated and characterized deletion mutations of Id2 and of Id4. Id2−/− mice exhibit abnormally rapid entrainment and an increase in the magnitude of the phase shift of the pacemaker. A significant proportion of mice also exhibit disrupted rhythms when maintained under constant darkness. Conversely, Id4−/− mice did not exhibit a noticeable circadian phenotype. In vitro studies using an mPer1 and an AVP promoter reporter revealed the potential for ID1, ID2, and ID3 proteins to interact with the canonical basic HLH clock proteins BMAL1 and CLOCK. These data suggest that the Id genes may be important for entrainment and operation of the mammalian circadian system, potentially acting through BMAL1 and CLOCK targets.

3 cell and dentritic cell populations [14], and regulation of B cell IgE class switching [15]. A defect in lactation associated with mammary gland development during pregnancy is associated with epithelial cell proliferation arrest [16]. Cardiac structural and functional conduction system abnormalities have been documented that result in left bundle branch block [17]. A partial penetrance has been described for abnormal development of the ureteropelvic junction resulting in hydronephrosis [18]. A reduced adiposity, and embryonic fibroblasts derived from these mice exhibit a diminished capacity for adipocyte differentiation [19].
Whilst we have not explored any specific studies of the immune system in our Id2 null mouse, we have observed considerable loss of animals during post-natal and early weaning phase of development. Loss of post-weaned mice is higher when mice are removed from a barrier facility, and we have found that supplementation with sterile food, and antibiotic treated water increases survivorship. These findings are consistent with a defect in the immune system of the Id2 null mice.
We have also been unable to breed successfully Id2 homozygote female with Id2 heterozygote male crosses, presumably due to the lactation defect, as identified in the Yokota Id2 knockout mouse.
Consistent with a low adiposity in the Yokota Id2 knockout mouse, we find that the Id2 null mice have a low level of fat storage and are lean (Hou, Watson, Israel and Duffield, unpublished data), contributing in part to the low body weight measurements.

Modulation of the photic response by ID2
Loss of ID2 has a clear effect on the photic response, and because the ID proteins can interact with and modify the activity of bHLH transcription factors, we reasoned that this influence might underlie the photic response phenotype. To test this we performed the experiments described in the Current Biology, Volume 19 4 text showing that ID1, ID2 and ID3 were each shown capable of blocking CLOCK:BMAL1-driven transcriptional activation with a potency comparable to mPER1, with ID2 showed the greatest inhibitory effect. The potent ability for ID proteins to interfere with the CLOCK:BMAL1 activation of clock gene and clock-controlled gene activity reveals a potential interplay between the ID HLH transcriptional inhibitors and the CLOCK and BMAL1 bHLH transcriptional activators in the circadian transcriptional-translational feedback loop. This would provide a novel manner in which additional components, namely Id genes, could modulate circadian function. In the context of CLOCK and BMAL1 proteins, which function as a heterodimer bound to DNA, the simplest interpretation of our results posits that ID proteins are binding directly to CLOCK and/or BMAL1 and in turn reducing the quantity of functional CLOCK:BMAL1 heterodimers available for binding to the E-box element.
This suggests that under appropriate circumstances, reduction or absence of ID2 could result in higher activity of CLOCK:BMAL1, in turn facilitating a stronger response to photic stimuli via transactivation of the mPer1and mPer2 gene promoters. As has been demonstrated, modulation of the quantity of functional CLOCK:BMAL1 heterodimer in the Clock mutant mouse results in changes in the induction profiles of the clock state variable genes, mPer1 and mPer2 [20,21].
These studies suggest that CLOCK:BMAL1 activity, whilst not serving actually to mediate the photic signal activating the period genes (this being pCREB [22][23][24]), can still modify the magnitude of the response to the photic stimulus. Additionally, a direct role for CLOCK in signaling to the period1 gene has also been proposed independent of the pCREB pathway: Ca2+dependent protein kinase C phosphorylation of CLOCK can regulate period 1 induction [25]. 7 Figure S1. Id2 is rhythmically expressed in rat-1 and mouse NIH3T3 fibroblasts 15 periodogram analysis, the ascending straight line represents a statistical significance of p = 0.001.
For Fourier analysis, a frequency of 0.042 cycles/hr corresponds to 1 cycle/24 hr, which is denoted by the arrow head below the chart where statistically significant (statistical significance was determined by the Clocklab program). All five mice in LD showed significant but in some cases extremely weak rhythmicity when measured from either passive or locomotor activity data. (C) The mean ± SEM magnitude of the phase shifts produced by light treatment is shown in the histogram for wild-type (white, n = 10), heterozygote (grey, n = 9) and mutant (orange, n = 10) mice. Extrapolated activity onsets of the first day following the 10 hr light treatment were used to determine the size of resultant phase delays. No significant difference was detected between genotypes (one-factor ANOVA, *p < 0.05).

Supplemental Experimental Procedures Cell treatment and RNA collection
Confluent Rat-1 and mouse NIH3T3 fibroblasts (ATCC, Manassas, VA) were treated with 50% horse serum, RNA harvested, and circadian patterns of gene expression analyzed by cDNA microarray and qRT-PCR, as detailed previously [4] and below, respectively. The start of serum treatment was designated time = 0 hr and gene expression was examined every 4-6 hr over a 48 hr period.

Mouse tissue collection
Mice (C3H/He wild-type [26]; light treatment initiated at CT16, or dark pulse controls collected between CT16.5 and CT17.5. The intervals between start of light stimulus and tissue collection was based on the profile of Id2 induction following mitogenic serum stimulation of fibroblasts [27], and the profiles of immediate early gene (e.g. c-fos) and clock gene (mPer1 and mPer2) induction in the SCN [28][29][30][31]. Brains were rapidly removed and snap frozen in isopentane at -60 °C on dry ice for 20 s before a 1mm coronal section was cut at the level of the optic chiasm using a fabricated twin blade cutter. A SCN punch from a frozen slice was then taken under a stereomicroscope using a flat-tipped 25G needle (internal diameter approximately 0.5 mm) and tissue stored on dry ice. The shape of the optic chiasm and third ventricle were used to define the SCN region. All six SCN punches were pooled to obtain sufficient RNA for subsequent procedures. The apical pole of the heart was dissected out and immediately placed into 0.5ml RNA later (Ambion, Austin, TX) and stored at 4 °C prior to RNA extraction.

Quantitative real-time RT-PCR analysis
Total RNA was prepared with Trizol reagent (Sigma-Aldrich, UK), DNAseI treated and cDNA synthesized as described previously [4,32]. SCN: one pooled sample at all time points; heart: 3 samples at each time point. Quantitative Real-time RT-PCR (qRT-PCR) was conducted using SYBR green and Taqman reagents as described previously [4,32]. Primers were designed for the  and samples were also run on a 3% agarose gel to confirm specificity by size of band. Expression data from mouse SCN and heart RNA were normalized to the expression of two internal controls, 18S rRNA and ARP. Expression data from mouse NIH3T3 cells and mouse embryonic fibroblasts was normalized to the expression of GAPDH. The Id2 Taqman primer and probe set was used with RNA samples from NIH3T3 cells, and from adult brain and mouse embryonic fibroblasts of Id2+/+ and Id2-/-mice. Data were initially analyzed using SDS 1.7 (Applied Biosystems). Relative mRNA abundance was calculated using two methods: raw data were exported as clipped files and analyzed as efficiency-corrected normalized expression as described previously [33], and as no statistical differences in amplification efficiency were observed, the mean efficiency was used for each transcript; and using the standard curve method outlined in the ABI Prism 7700 Sequence Detection System User Bulletin #2 (Perkin Elmer/Applied Biosystems) [4]. Significant differences were determined by one-factor ANOVA, followed by post-hoc Dunnett's t-tests (p < 0.05).

Generation of Id2 mutant mice
A replacement vector was designed that deletes two exons that include the entire coding region of mouse ID2 protein. The targeted allele was obtained in embryonic stem cells ( Figure S2B) and product. Homozygous mutants are morphologically dissimilar to their wild-type littermate controls, being on average 27% smaller (F 2,56 = 9.1, p = 0.0004), and our other broad observations of the Id2-/-mice are consistent with the Id2 null mice generated by Yokota and colleagues [12]. Histological analysis of SCN coronal sections showed no gross anatomical difference between Id2+/+ and Id2-/-F2 littermates (Figure S3), indicating that the circadian phenotypes in the mutants are not due to a gross developmental defect in the basic organization of the SCN.
We isolated a genomic clone from a mouse 129/sv genomic P1 phage library (Genomic Systems, USA) using a 1.3 kb full length Id2 complementary DNA probe [34]. Chimeric males were bred to C57BL/6J females, and F1 mice interbred to produce F2 mice.
Homozygous mice were produced through heterozygous intercrosses. Id2-deficient mice were initially established on a C57BL/6J background, but for breeding purposes due to extremely low production of Id2-/-homozygotes (< 1% survivorship from heterozygote x heterozygote crosses), transgenic mice were later established in a mixed genetic background (129sv/C57BL6J/FBVN; resulting in ~10% survivorship from heterozygote x heterozygote crosses). Homozygote wild-type and heterozygote littermates were used as control mice in all subsequent experiments, as recommended [35].
Genotypes were determined by Southern blot analysis and/or PCR of tail biopsy DNA ( Figure S2). Southern blot analysis was performed as described above. PCR primers specific for the wild-type and targeted alleles (neo) that were used (Id2-wild-type intron forward 5'-AGG CGC 29 Targeted deletion of the Id2 transcript was also confirmed by qRT-PCR from mRNA extracted from adult brain and liver, and from mouse embryonic fibroblasts of Id2+/+, Id2+/-and Id2-/-mice (liver mRNA is shown in Figure S2 D). Total RNA was isolated from tissues using Trizol reagent and qRT-PCR applied as described above using Taqman primers and probe specific for the detection of the Id2 cDNA upstream of the exon 3 region (i.e. within exon 2). The 2.2 kb mutant protein product of Neo-Exon3 contains no known functional domains as Exon 3 is wholly 3' UTR and contains no coding sequence, and thus is unlikely to form any functional protein. Absence of ID2 protein was confirmed by Western blot analysis of liver samples harvested at CT12 ( Figure   S2 E). These gene, mRNA and protein analyses, and the lack of any phenotype in the Id2+/-mice strongly suggest that our targeting strategy generated a null allele at Id2.

Western blots
Liver was homogenized in RIPA lysis buffer, plus 1% protease/phosphatase inhibitor cocktail

Generation of Id4-/-mice
Id4 mutant mice were generated by homologous recombination as described in Yun et al. [36]: Exons 1 and 2 were replaced by the coding regions of GFP and the neomycin-resistance genes.
Following transfection into embryonic stem cells, recombination events were confirmed by Southern blot (6.5 band) and PCR analyses. PCR analysis of mouse tails confirmed genotype of individual mice. Mice were established and maintained on a C57BL/6J background, and unlike the Id2-/-mice showed normal Id4-/-genotype productivity with heterozygote x heterozygote crosses. SPOT software program (Diagnostic Instruments) and the mean of these values calculated.

Histological analysis of
Significant differences were determined by one-factor ANOVA (with False Discovery Rate Correction where interactions were apparent).
Locomotor activity monitoring and circadian phenotype analysis 33 and entrained to a 12:12 LD for at least 14 days, and transferred to constant light for 30 days.
Fourier and ! 2 periodogram analyses were undertaken on the third 10 day duration in LL. All other behavioral data shown was produced from a single population of mice studied at Dartmouth Medical School. Significant differences were determined by Fisher's Exact test (p < 0.05), Student's t-test (two-tailed distribution, p < 0.05), paired t-test (p < 0.05), and one-factor ANOVA, followed by post-hoc Dunnett's t-tests (p < 0.05). Unless stated otherwise, the statistical test refers to a one-factor ANOVA.